The cell wall is a dynamic structural barrier that can determine the outcome of the interactions between plants and pathogens. At the sites of interaction, plants actively reinforce the cell wall though development of cell wall appositions, called papillae (Zeyen et al., 2002). Papillae formation appears to be a consequence of pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and/or damage-associated molecular pattern (DAMP)-triggered immunity (DTI), that is observed during many plant-pathogen interactions (Boller and Felix, 2009; Malinovsky et al., 2014). Although, the role of papillae is not fully understood, it is believed that they provide a physical and chemical barrier that can completely prevent pathogen entry or at least delay pathogen penetration into the cells so that other defense strategies can be activated (Stone and Clarke, 1992; Huckelhoven, 2005). Without successful penetration the pathogen is unable to develop a feeding structure called the haustorium which is required to absorb water and nutrients from the plant (Bélanger et al., 2002). The composition of papillae may vary among different species, but commonly found papillae components are cell wall polysaccharides, cell wall proteins, reactive oxygen species, phenolics and anti-microbial compounds (Zeyen et al., 2002; Hückelhoven, 2014). Among the cell wall polysaccharides, callose has commonly been assumed to be the major polysaccharide in the papillae that form during plant/pathogen interactions, but it has recently been shown that arabinoxylan and cellulose are also major constituents of papillae that form during infection of barley by the powdery mildew pathogen Blumeria graminis f. sp. hordei (Bgh) (Chowdhury et al., 2014). Earlier and heavier deposition of these polysaccharides are proposed to aid in arresting the progress of the fungal penetration peg through effective papillae formation (Chowdhury et al., 2014). A glucan synthase gene (HvGsl6) and a cellulose synthase-like D2 gene (HvCslD2) have been shown to contribute toward the biosynthesis of callose and cellulose in barley papillae, respectively, and down regulation of these genes results in increased penetration susceptibility to Bgh (Chowdhury et al., 2016; Douchkov et al., 2016). Together these results suggest that polysaccharide deposition in papillae is essential for the mechanism of penetration resistance in barley against powdery mildew infection.

Arabinoxylan was detected in barley papillae using the LM11 monoclonal antibody, which binds to an epitope of low substituted xylan, whereas no significant labeling was observed using the LM10 antibody, which binds unsubstituted xylan (Chowdhury et al., 2014). The arabinoxylan epitopes were distributed throughout the entire papillae, in which two zones could be readily recognized. An inner core consisted of arabinoxylan, callose and phenolics, while an outer layer, or coat, contained arabinoxylan and cellulose. As the LM11 antibody binds to wheat arabinoxylan with a low degree of arabinose substitutions (McCartney et al., 2005), we can assume that some degree of substitution of the backbone xylan with arabinosyl residues is likely to be present in barley papillae. However, we cannot be sure what other substitutions may occur along the xylan backbone and will therefore refer to the polysaccharide as heteroxylan.

Cell wall reinforcement in response to biotrophic and necrotrophic pathogen attack can be mediated through oxidative cross-linking between polysaccharides, structural proteins and phenolic compounds (Passardi et al., 2004; Deepak et al., 2010). The presence of heteroxylan in papillae provides an opportunity for increased cell wall polymer cross-linking through associated phenolic compounds, such as ferulic acid. More highly cross-linked wall polysaccharides would presumably toughen the wall and increase resistance against fungal penetration, whether that penetration is mediated through physical pressure or by enzymatic hydrolysis. In monocotyledonous plants, ferulic acid can be ester-linked to arabinofuranosyl substituents of the xylan backbone. The ferulic acid may undergo oxidative dimerization to covalently link adjacent feruloylated arabinoxylan chains or, alternatively, the heteroxylan chains might be cross-linked with lignin (Burr and Fry, 2009; Marcia, 2009). There is evidence that arabinoxylan with low degrees of substitution can interact with cellulose microfibrils and other polymers via extensive intermolecular hydrogen bonding (Busse-Wicher et al., 2014) and this is also likely to influence the strength and elasticity of walls, and contribute to the resistance of the walls to enzymatic degradation. It can be argued that heteroxylan, as a major component of papillae, helps to provide mechanical strength against the fungal penetration process. Thus, genetic manipulation of heteroxylan content or structure may impact upon the plant's susceptibility to infection.

The complete process of heteroxylan biosynthesis in the plant cell wall is not fully understood. However, there is evidence to suggest that the assembly of the xylan backbone and its substituents involves at least six families of glycosyltransferases (GTs) found in the Carbohydrate-Active EnZymes (CAZy) database (Cantarel et al., 2009), namely GT8, GT31, GT43, GT47, GT61, and GT75 (Rennie and Scheller, 2014). The predicted role for each of these GT families in assembling heteroxylan in grasses is presented in Figure 1. Two members of the GT43 family, known as Irregular Xylem 9 (IRX9) and IRX14, and a member of the GT47 family, IRX10, have putative xylosyltransferase activities required for synthesizing the xylan backbone (Brown et al., 2007; Lee et al., 2007; Wu et al., 2009; Mortimer et al., 2015). Two members of the GT8 family, termed Glucuronic Acid Substitution of Xylan (GUX) 1 and GUX2, introduce glucuronyl (GlcA) substituents onto the xylan backbone (Mortimer et al., 2010; Rennie et al., 2012). Members of the GT75 family, which encode UDP-arabinose mutases (UAM) or Reversibly Glycosylated Protein (RGP), are responsible for converting UDP-arabinopyranose to UDP-arabinofuranose, which is the substrate for xylan arabinosyltransferases (Rautengarten et al., 2011; Hsieh et al., 2016). Some members of the GT61 family have xylan arabinosyltransferase activity that adds arabinosyl residues onto the xylan backbone (Anders et al., 2012), while another member of the GT61 family is capable of mediating xylosyl substitution of arabinosyl residues on the xylan backbone (Chiniquy et al., 2012). The GT31 family was also included in our study due to its reported association with heteroxylan synthesis, although there is no direct evidence that these enzymes are directly involved. Several GT31 genes encode galactosyltransferases that transfer galactosyl residues from UDP-galactose to arabinogalactan-proteins (AGPs) (Nguema-Ona et al., 2015). AGPs may be cross-linked with heteroxylan (Tan et al., 2013) and may also play a role in pathogen defense responses (Nguema-Ona et al., 2013).

In the current study, we have investigated the potential roles in penetration resistance against Bgh of genes that have been implicated in heteroxylan biosynthesis. Up- or down-regulation of transcript levels of selected candidate genes during infection indicated that the susceptibility against Bgh penetration is significantly altered in barley epidermal cells. The identification of these genes might provide new opportunities for the generation of novel crop lines with greater disease resistance.

The dynamics of heteroxylan biosynthesis represents an area of untapped potential for activating or enhancing cell wall associated plant defense mechanisms. Heteroxylan is an important component of barley papillae and is associated with penetration resistance against powdery mildew pathogens (Chowdhury et al., 2014). The precise structure of heteroxylans in papillae is not known, because it is technically difficult to isolate individual papillae for chemical analysis. However, the heteroxylans in cell walls of grasses and cereals has been shown to contain a (1,4)-β-xylan backbone and, depending upon the species and tissue type, the backbone is substituted to varying degrees with α-arabinofuranosyl (Araf) residues, α-glucuronosyl residues (GlcA), and with feruloylated arabinofuranosyl residues (Carpita and Gibeaut, 1993; Stone and Fincher, 2004; Ebringerová et al., 2005; Burton and Fincher, 2012). Due to the inherent capability of feruloylated heteroxylans to form covalent cross-links via their phenolic acid residues not only with other arabinoxylan molecules, but also with other wall polymers, one might expect that this cross-linking could lead to an increase in mechanical strength of the walls and that the presence of feruloylated heteroxylans in barley papillae could be of great importance in plant resistance to fungal penetration.

To understand the role of heteroxylan in papillae-based cell wall reinforcement, we examined candidate genes selected from various glycosyltransferase (GT) families that have been implicated in arabinoxylan biosynthesis (Rennie and Scheller, 2014). A number of candidate genes from heteroxylan-associated GT families were significantly up-regulated in the barley epidermal layer during powdery mildew infection, at times corresponding to the period when heteroxylan was being synthesized in the papillae (Figure 2). Similar expression profiles were observed during infection by the non-adapted pathogen Bgt, suggesting that these genes are somehow involved in basal plant defense mechanisms that are not the ultimate determinants of successful penetration or resistance outcomes.

In this study, silencing and over-expression of several GT43 and GT47 candidate genes resulted in susceptibility phenotypes that might be predicted from the recent discovery that heteroxylans are present in barley papillae (Chowdhury et al., 2014). However, for all candidate genes, the predicted reciprocal resistance and susceptibility phenotypes were not obtained by up-regulation and down-regulation of their expression levels. The GT43 gene (MLOC_54026) is the barley homolog of the rice GT43E gene (LOC_Os05g48600, Figure S1), which has been demonstrated to complement the irregular xylem mutation (IRX9) in Arabidopsis (Lee et al., 2014). Mutation of the Arabidopsis IRX9 gene (At2g37090) results in defective stem mechanical strength, vessel morphology, xylan content, GlcA side chains, xylan chain length, and xylosyltransferase activity (Lee et al., 2010; Mortimer et al., 2015). Biochemical assays for xylan xylotransferase activity demonstrated that the IRX9 protein is capable of elongation of the xylan backbone in the presence of another GT43 cofactor IRX14 (Lee et al., 2012). We have found that over-expression of this GT43 gene (MLOC_54026) in combination with members of the GT47 family leads to significant increases in resistance. The contrasting level of resistance obtained from individually over-expressing the GT43 (MLOC_54026) and GT47 (MLOC_5743) genes compared to when they are over-expressed in combination suggests that these two gene family members may somehow co-operate during the synthesis of the xylan backbone in effective papillae. This result is supported by reports that GT43 and GT47 enzymes might co-exist in a single xylan synthase complex and might need to interact for effective heteroxylan biosynthesis (Zeng et al., 2010). Although, the current data suggest that xylan represents a significant barrier to fungal penetration during the plant's defense responses, the direct impacts on disease resistance and the precise molecular mechanisms in defense-related cell walls that would explain the effects of silencing or over-expressing the GT43 and GT47 genes have yet to be defined. However, silencing of specific GT43 and GT47 gene family members can significantly improve wall digestibility by cell wall degrading enzymes in grasses and eudicots (Lee et al., 2009; Petersen et al., 2012). These results are consistent with the data presented here, which support the hypothesis that GT43 and GT47 enzymes, through their xylan biosynthesis activity, increase cell wall rigidity and hence resistance against enzymatic degradation by pathogens. It should also be pointed out that the synthesis of heteroxylans with lower degrees of substitution and/or the removal of some of the arabinosyl residues by α-arabinofuranosidases would be expected to facilitate intermolecular binding, as would an increase in ferulic acid cross-links. In both cases one might reasonably expect that the papillae would be “toughened” and would be more resistant to fungal penetration.

In addition to the GT43 and GT47 combinations, five other candidate genes produced significant resistance phenotypes when their expression levels were up-regulated. Within this group of genes were representatives of the GT8, GT61, and GT75 families. Silencing of two GT75 members (MLOC_64204 and MLOC_6065) had no significant effect on pathogen susceptibility levels, possibly due to redundancy. However, over-expression of these two genes individually led to a significant increase in resistance. A recent study has shown that both MLOC_64204 and MLOC_6065 have UDP-arabinose mutase (UAM) activity and are able to convert UDP-l-arabinopyranose to UDP-l-arabinofuranose (Hsieh et al., 2016). UDP-l-Arabinofuranose is the required form of substrate arabinose for arabinosyltransferases that attach arabinosyl residues to the xylan backbone. Given that these two genes encode enzymes that are presumed to perform the same enzymatic function, it is not unexpected that silencing single members of the family had no effect. Information on silencing mutase genes and the impact on disease resistance is lacking. However, a recent study shows that in Brachypodium, silencing of GT75 genes increases xylanase mediated digestibility of the cell wall by two-fold and is related to a significant decrease in cell wall ferulic acid and p-coumaric acid concentrations (Rancour et al., 2015).

As shown in a phylogeny of the GT8 family, the next two candidates, MLOC_19204 and MLOC_65693, are present in a cluster that contains Arabidopsis GT8 proteins (Figure S5) that have been shown to possess glucuronosyltransferase activity (Rennie et al., 2012) suggesting that they may be involved in adding glucuronyl residues to the xylan backbone (Figure 1). Members of the GT8 family have also been implicated in pectin biosynthesis (Mohnen, 2008), but levels of pectic polysaccharides in the walls of barley and other grasses are generally low (Stone and Fincher, 2004). As noted above for arabinosyl substitution, removal of glucuronyl substituents could also alter intermolecular interactions and enhance the strength of the walls and the papillae in such a way that the attack on cell walls by pathogens is inhibited (Mortimer et al., 2010). Although, a direct relationship between glucuronyl residues on the xylan backbone and disease resistance has not been demonstrated, the absence of glucuronyl residues in the xylan backbone achieved through double mutation of the GUX genes (GT8 members) results in weaker stem and easier cell wall digestibility with cell wall degrading enzymes, compared to the wild type (Mortimer et al., 2010). This result suggests that glucuronyl residues on heteroxylans may be required for improving the overall strength of the wall against pathogen attack, by limiting enzymatic digestibility by the pathogen.

Another candidate gene that led to an increase in penetration resistance was MLOC_64310, which belongs to the GT61 family. This gene belongs to a monocotyledon-specific phylogenetic clade that consists largely of uncharacterised genes and forms a sister clade to the Arabidopsis genes AT3G18170 and AT3G18180 (Figure S3). The importance of GT61 family in penetration resistance is strengthened further by enhanced RSI upon TIGS of two additional GT61-family members. The fact that no reciprocal infection phenotypes were observed by TIGS or over-expression of individual GT61 genes again suggests the existence of functional redundancy within the family. Several members of the GT61 gene family are likely to encode arabinofuranosyl transferases, which add arabinosyl residues to the (1,4)-β-xylan backbone (Anders et al., 2012). In grass species, the arabinosyl residues can be covalently linked to ferulic acid residues, which may undergo oxidative dimerization reactions and become covalently attached to adjacent feruloylated heteroxylan chains (Burr and Fry, 2009; Marcia, 2009). Modified susceptibility levels achieved by altering gene expression levels may be attributable to differential cross-linking of heteroxylan with other polysaccharides and phenolic compounds, since previous studies showed that in grasses, arabinosyl-ferulic acid crosslinking also improved resistance against pathogen fungi by limiting enzymatic digestibility of the modified wall (Bily et al., 2003; Santiago and Malvar, 2010; Lionetti et al., 2015).

The central goal of the current study was to identify candidate genes implicated in heteroxylan biosynthesis during powdery mildew infection that might have significant impacts on penetration resistance. The main challenge for characterizing the role of individual genes involved in heteroxylan biosynthesis in papillae is embodied in the presence of multiple genes with overlapping function within each GT family and the associated interpretative constraints. Other members of a particular family might potentially compensate in a biosynthetic process when the candidate gene is silenced. Thus, the effects of over-expression of a single gene might be masked by altered expression patterns of related genes or of other genes that might need to be co-expressed to form a functional multi-enzyme complex.

Houston et al. (2016) identified that the GT8, GT61, and GT75 families are upregulated on average across abiotic and biotic stresses in Arabidopsis and barley. Our results suggest that at least one member of each of the gene families associated with arabinoxylan biosynthesis play vital roles in plant-pathogen resistance mechanisms. This is presumably related to modified heteroxylan structures and/or altered levels of heteroxylan accumulation in papillae, both of which might provide greater mechanical strength and resistance against the physical and enzymatic mechanisms of fungal penetration. The cell wall composition and integrity could also play a role in the activation of the defense response through DAMP-signaling and pattern-triggered immunity (Malinovsky et al., 2014). If heteroxylan fragments were able to activate these pathways it would be possible to conceive that small changes in polysaccharide structure could result in the oligosaccharides released during pathogenesis not being detected by cell surface receptors.

Due to technical difficulties in labeling arabinoxylan with a range of specific antibodies in the transiently transformed cells, a direct correlation of gene expression levels and arabinoxylan biosynthesis in papillae was not possible. In order to confirm this link, further experiments will be focused on demonstrating that there is a direct link between the candidate genes described here and the biosynthesis of heteroxylan. Stable transgenic lines expressing the selected genes of interest will allow immunohistological characterization of papillae and cell wall heteroxylan, however, this will be limited to the antibodies available for different heteroxylan structures and may not provide the level of detail required to completely elucidate their role in penetration resistance. Expression and purification of the candidate genes in heterologous systems could allow in vitro biochemical analyses to be performed, again to demonstrate the synthase activity of each candidate gene. Nevertheless, data generated in the present study is consistent with a role for heteroxylan biosynthesis genes in plant pathogenesis and suggests that manipulation of the amounts and fine structure of heteroxylan has considerable potential for the development of disease resistant lines in the future.

GF, RB, PS, and AL planned and designed the research. JC, JS, NS, SL, JR, DD, and AL performed the experiments and analyzed the data. JC, PS, GF, RB, and AL wrote the manuscript.

The work was supported by grants from the Australian Research Council (to GF and RB) and from the German Ministry of Education and Research (BARLEY-fortress, to PS).